In spite of numerous advances in medical research, cancer remains a leading cause of death throughout the developed world. Non-specific approaches to cancer management, such as surgery, radiotherapy and generalized chemotherapy, have been successful in the management of a selective group of circulating and slow-growing solid cancers. However, many solid tumors are considerably resistant to such approaches, and the prognosis in such cases is correspondingly grave.
One example is brain cancer. Each year, approximately 15,000 cases of high grade astrocytomas are diagnosed in the United States. The number is growing in both pediatric and adult populations. Standard treatments include cytoreductive surgery followed by radiation therapy or chemotherapy. There is no cure, and virtually all patients ultimately succumb to recurrent or progressive disease. The overall survival for grade IV astrocytomas (glioblastoma multiforme) is poor, with .about.50% of patients dying in the first year after diagnosis. Because these tumors are aggressive and highly resistant to standard treatments, new therapies are needed.
An emerging area of cancer treatment is immunotherapy. The general principle is to confer upon the subject being treated an ability to mount what is in effect a rejection response, specifically against the malignant cells. There are a number of immunological strategies under development, including: 1. Adoptive immunotherapy using stimulated autologous cells of various kinds; 2. Systemic transfer of allogeneic lymphocytes; 3. Intra-tumor implantation of immunologically reactive cells; and 4. Vaccination at a distant site to generate a systemic tumor-specific immune response.
The first of the strategies listed above, adoptive immunotherapy, is directed towards providing the patient with a level of enhanced immunity by stimulating cells ex vivo, and then readministering them to the patient. The cells are histocompatible with the subject, and are generally obtained from a previous autologous donation.
One approach is to stimulate autologous lymphocytes ex vivo with tumor-associated antigen to make them tumor-specific. Zarling et al. (1978) Nature 274:269-71 generated cytotoxic lymphocytes in vitro against autologous human leukemia cells. Lee et al. (1996) abstract, Gastroenterology conducted an in vitro mixed lymphocyte culture with inactivated leukemic blast cells and autologous lymphocytes, and generated effector T lymphocytes cytotoxic for a tumor antigen on autologous blast cells. An MHC D-locus incompatibility was thought to be necessary to provide proper help in the lymphocyte culture. Lesham et al. (1984) Cancer Immunol Immunother. 17:117-23 developed cytotoxic responses in vitro against murine thymoma cells by allosensitization.
Gately et al. (1982) J. Natl. Cancer Inst. 69:1245-54 found that 5 out of 9 human glioma cell lines did not elicit allogeneic cytolytic lymphocyte responses in ex vivo cultures. However, if inactivated, allogeneic lymphocytes were provided as stimulator cells in the cultures, tumor-specific cytolytic T lymphocytes and non-specific non-T effectors were generated to 4 of the nonstimulatory lines. In U.S. Pat. No. 5,192,537, Osband suggests activating a tumor patient's mononuclear cells by culturing them ex vivo in the presence of tumor cell extract and a non-specific activator like phytohemagglutinin or IL-1, and then treating the culture to deplete suppresser cell activity.
Despite these experimental observations, systemic administration of ex vivo-stimulated autologous tumor-specific lymphocytes has not become part of standard cancer therapy.
Autologous lymphocytes and killer cells may also be stimulated non-specifically. In one example, Fc receptor expressing leukocytes that can mediate an antibody-dependent cell-mediated cytotoxicity reaction are generated by culturing with a combination of IL-2 and IFN-.gamma. (U.S. Pat. No. 5,308,626). In another example, peripheral blood-derived lymphocytes cultured in IL-2 form lymphokine-activated killer (LAK) cells, which are cytolytic towards a wide range of neoplastic cells, but not normal cells. LAK are primarily derived from natural killer cells expressing the CD56 antigen, but not CD3. Such cells can be purified from peripheral blood leukocytes by IL-2-induced adherence to plastic (A-LAK cells; see U.S. Pat. No. 5,057,423). In combination with high dose IL-2, LAK cells have had some success in the treatment of metastatic human melanoma and renal cell carcinoma. Rosenberg (1987) New Engl. J Med. 316:889-897. This strategy is labor-intensive, costly, and not suited to all patients. Schwartz et al. (1989) Cancer Res. 49:1441-1446 showed that A-LAK cells are superior to LAK cells at reducing lung and liver metastases of breast cancer in experimental animal models, but this was not curative and there were no long-term survivors.
For examples of trials conducted using LAK in the treatment of brain tumors, see Merchant et al. (1988) Cancer 62:665-671 & (1990) J. Neuro-Oncol. 8:173-198; Yoshida et al. (1988) Cancer Res. 48:5011-5016; Barba et al. (1989) J. Neurosurg. 70:175-182; Hayes et al. (1988) Lymphokine Res. 7:337-345; and Naganuma et al (1989) Acta Neurochir. (Wien) 99:157-160. Another study proposes therapy for recurrent high-grade glioma using autologous mitogen-activated and IL-2 stimulated (MAK) killer lymphocytes, in combination with IL-2. Jeffes et al. (1991) Lymphokine Res. 10:89-94. While none of these trials was associated with serious clinical complications, efficacy was only anecdotal or transient. Induction of tumor-specific immunity in patients receiving such treatments has not been shown.
Another form of adoptive therapy using autologous cells has been proposed based on observations with tumor-infiltrating lymphocytes (TIL). TILs are obtained by collecting lymphocyte populations infiltrating into tumors, and culturing them ex vivo with IL-2. Finke et al. (1990) Cancer Res. 50:2363-2370 have characterized cytolytic activity of CD4+ and CD8+ TIL in human renal cell carcinoma. TILs have activity and tumor specificity superior to LAK cells, and have been experimentally administered, for example, to humans with advanced melanoma. Rosenberg et al. (1990) New Engl. J. Med. 323:570-578. The effector population within TILs may be cytotoxic T lymphocytes (CTL) which are primed to be tumor-specific in the host and are devoid of lytic granules, and become transformed into cytolytic lymphoblasts when stimulated in culture. Berke et al.(1988) J. Immunol. 129:303 ff. Unfortunately, TILs can only be prepared in sufficient quantity to be clinically relevant in a limited number of tumor types. These strategies remain experimental, especially in human therapy.
The second of the strategies for cancer immunotherapy listed earlier is adoptive transfer of allogeneic lymphocytes. The rationale of this experimental strategy is to create a general level of immune stimulation, and thereby overcome the anergy that prevents the host's immune system from rejecting the tumor. Strausser et al. (1981) J. Immunol. Vol. 127, No. 1 describe the lysis of human solid tumors by autologous cells sensitized in vitro to alloantigens. Zarling et al. (1978) Nature 274:269-71 demonstrated human anti-lymphoma responses in vivo following sensitization with allogeneic leukocytes. Kondo et al. (1984) Med Hypotheses 15:241-77 observed objective responses of this strategy in 20-30% of patients, and attributed the effect to depletion of suppressor T cells. The studies were performed on patients with disseminated or circulating disease. Even though these initial experiments were conducted over a decade ago, the strategy has not gained general acceptance, especially for the treatment of solid tumors.
The third of the immunotherapy strategies listed earlier is intra-tumor implantation. This is a strategy directed at delivering effector cells directly to the site of action. Since the transplanted cells do not circulate, they need not be histocompatible with the host. Intratumor implantation of allogeneic cells may promote the ability of the transplanted cells to react with the tumor, and initiate a potent graft versus tumor response.
Kruse et al. (1990) Proc. Natl. Acad. Sci U.S.A. 87:9577-9581 demonstrated that direct intratumoral implantation of allogeneic cytotoxic T lymphocytes (CTL) into brain tumors growing in Fischer rats resulted in a significant survival advantage over other populations of lymphocytes, including syngeneic CTL, LAK cells, adherent-LAK cells or IL-2 alone. Redd et al. (1992) Cancer Immunol. Immunother. 34:349-354 developed cytotoxic T lymphocytes specific for an allogeneic brain tumor in rats. The lymphocytes were specific for a determinant expressed only by the tumor, and were predicted to be useful for therapeutic purposes in vivo. Kruse et al. (1994) J. Neurooncol. 19:161-168 prepared CTLs from four MHC incompatible rat strains, and used them to treat Fischer rats bearing established 9L brain tumors. CTL were administered on a biweekly schedule, a different MHC incompatible CTL preparation being administered each time. Animals without tumor showed minimal localized brain damage. Those with tumors either showed: a) mononuclear cell infiltration, massive tumor necrosis beginning 2-4 days after treatment, and total tumor destruction by 15 days; or b) cellular infiltration, early tumor destruction, and then tumor regrowth progressing to death of the animal. Tumor regressor animals were resistant to intracranial rechallenge with viable tumor cells. Kruse et al. (1994). Intratumor CTL implants may optionally be combined with chemotherapy using cyclophosphamide. Kruse et al. (1993) J. Neurooncol. 15:97-112.
Despite the promise of intratumor implantation techniques, several caveats remain. For one thing, implantation is frequently performed by surgical techniques, which may be too invasive for routine maintenance. In addition, the strategy is directed at generating a local response, not at generating a systemic response that is generally necessary for protection against metastases.
The fourth of the immunotherapy strategies listed earlier is the generation of an active systemic tumor-specific immune response of host origin. The response is elicited from the subject's own immune system by administering a vaccine composition at a site distant from the tumor. The specific antibodies or immune cells elicited in the host as a result will hopefully migrate to the tumor, and then eradicate the cancer cells, wherever they are in the body.
Various types of vaccines have been proposed, including isolated tumor-antigen vaccines and anti-idiotype vaccines. Mitchell et al. (1993) Ann. N.Y. Acad. Sci. 690:153-166 have treated cancer patients with mechanical lysates from a plurality of allogeneic melanoma cell lines, combined with the adjuvant DETOX.TM.. These approaches are all based on the premise that tumors of related tissue type all share a common tumor-associated antigen. For patients with tumors that did lot acquire expression of the antigen during malignant transformation, or that subsequently differentiated so as not to express it, none of these vaccines will be successful.
An alternative approach to an anti-tumor vaccine is to use tumor cells from the subject to be treated, or a derivative of such cells. For review see, Schirrmacher et al. (1995) J. Cancer Res. Clin. Oncol. 121:487-489. In U.S. Pat. No. 5,484,596, Hanna Jr. et al. claim a method for treating a resectable carcinoma to prevent recurrence or metastases, comprising surgically removing the tumor, dispersing the cells with collagenase, irradiating the cells, and vaccinating the patient with at least three consecutive doses of about 10.sup.7 cells. The cells may optionally be cryopreserved, and the immune system may be monitored by skin testing. This approach does not solve the well-established observations that many tumors are not naturally immunogenic. Many patients from which tumors have been resected are either tolerant or unable to respond to their own tumor antigen, even when comprised in a vaccine preparation.
Several ways of preparing autologous or syngeneic tumor cells have emerged that potentially enhance immunogenicity. Tumor cells may be combined with extracts of bacillus Calmette-Guerin (BCG) or the A60 mycobacterial antigen complex, or mixed with vaccinia virus or Newcastle Disease Virus (NDV). Guo and coworkers (WO 95/16775) suggested that tumor cells be fused with membrane components of a second cell such as a B cell that has a greater immunogenic potential.
In yet another approach, autologous or syngeneic tumor cells are genetically altered to produce a costimulatory molecule. Examples of costimulatory molecules include cell surface receptors, such as the B7-1 costimulatory molecule or allogeneic histocompatibility antigens. Other examples are secreted activators, including cytokines. For reviews see, Pardoll et al. (1992) Curr. Opin. Immunol. 4:619-23; Saito et al. (1994) Cancer Res. 54:3516-3520; Vieweg et al.(1994) Cancer Res. 54:1760-1765; Gastl et al. (1992) Cancer Res. 52:6229-6236; and WO 96/07433). Tumor cells have been genetically altered to produce TNF-.alpha., IL-1, IL-2, IL-3, IL-4, IL-6, IL-7, IL-10, IFN-.alpha., IFN-.gamma. and GM-CSF. See, for example, Santin et al. Gynecol. Oncol. 58:230-239, 1995b; Int. J. Gynecol. Cancer 5:401-410, 1995c; Am. J. Obst. Gynecol. 174:633-639, 1996.
Golumbek et al. (1989) reported that mouse renal carcinoma cells inserted with a gene for IL-4 was strongly immunogeneic for systemic T cell immunity, and protected mice against a subsequent lethal challenge with unmodified, parental tumor cells. Cavallo et al. 1991 & 1992. Antitumor immunity is intensified by a cancer vaccine that produces both GM-CSF and IL-4. Wakimoto et al. (1996) Cancer Res. 56:1828-33. The cytokine or cytokine combination may recruit or stimulate cells of the immune system, and thereby overcome the normal barrier to immunity. Certain cytokines also affect the expression of major histocompatibility molecules and intercellular adhesion molecules by cancer cells (Santin et al. 1995a, Int. J. Cancer 65:688-694; Santin (1996) Am. J. Obst. Gynecol), potentially improving immunogenicity.
The experiments with cytokine-secreting histocompatible tumor cells have been done chiefly in genetically restricted animal models, which are not directly equivalent to a heterogeneous human patent population. Colombo et al. (1995) Cancer Immunol. Immunother. 41:265-270. Not all cancer types are responsive to the same cytokines. There are concerns about injecting human patients with replication-competent tumor cells, particularly after genetic alteration. In addition, there is usually not enough time to genetically alter and grow up sufficient cells of the patient to be treated for use in a vaccine.
Blumbach (WO 96/05866) has suggested vaccines of live tumor cells transduced with: a) a gene coding for an immunostimulatory protein; b) a cytokine; and c) a thymidine kinase gene. The composition is provided as live cells which can grow in vivo and stimulate a response, and then be selectively killed via the thymidine kinase. The possibility of escape mutants is likely to be a subject of regulatory concern for this approach in human therapy. Golumbek et al. (1992) J. Immunother. 12:224-230 have shown that proliferating tumor cells with suicide genes can also survive toxin treatment when they exit the cell cycle temporarily or are sequestered pharmacologically.
As an alternative, Cohen (WO 95/31107) suggested that neoplastic disease can be treated with a cellular immunogen comprising allogeneic cells genetically altered to express one or more cytokines, and also to express tumor-associated antigens encoded by autologous genomic tumor DNA. In this approach, an allogeneic cell (exemplified as a mouse LM cell) is genetically altered to express: a) a cytokine; and b) a tumor-associated antigen autologous to the subject to be treated. Accordingly, the vaccine need not comprise live tumor cells.
However, application of the Cohen invention to human subjects would require prior knowledge for each patient of a particular tumor-associated antigens expressed by the particular tumor. Many human cancers of widespread clinical interest do not have reliable commonly-shared markers. Once a relevant marker is identified for a particular patient, a cell line must be engineered accordingly, and cultured to the required density prior to treatment. Thus, each patient would become their own research and development project. Since the immune response would be focused only at the particular tumor-associated antigen used, it may be less effective than one directed against the spectrum of antigen expressed by a complete tumor cell. Furthermore, the vaccine comprises a live genetically altered cell line, raising the concerns outlined earlier. Cohen demonstrated only a modest improvement in survival in the animal studies, and failed to provide any evidence that his formulation would be effective in human cancer patients.
A suitable strategy for a human anti-tumor cellular vaccine has to contend with the following problems: a) heterogeneity amongst tumors (even tumors of the same type) in the display of tumor-associated antigens; b) heterogeneity in the immune response between individuals with regards to both antigens and cytokines; c) ethical and regulatory concerns about compositions that may be used in humans; and d) lack of development time in most clinical settings, limiting the ability to engineer new cell lines or otherwise tailor the vaccine to each patient.